4998 J. Agric. Food Chem.

2002, 50, 4998−5006

Analysis of Acrylamide, a Carcinogen Formed in Heated

Foodstuffs
EDEN TAREKE,† PER RYDBERG,† PATRIK KARLSSON,‡ SUNE ERIKSSON,‡ AND
MARGARETA TÖRNQVIST*,†
Department of Environmental Chemistry, Stockholm University, S-106 91 Stockholm, Sweden, and
AnalyCen Nordic AB, Box 905, S-531 19 Lidköping, Sweden

Reaction products (adducts) of acrylamide with N termini of hemoglobin (Hb) are regularly observed
in persons without known exposure. The average Hb adduct level measured in Swedish adults is
preliminarily estimated to correspond to a daily intake approaching 100 µg of acrylamide. Because
this uptake rate could be associated with a considerable cancer risk, it was considered important to
identify its origin. It was hypothesized that acrylamide was formed at elevated temperatures in cooking,
which was indicated in earlier studies of rats fed fried animal feed. This paper reports the analysis of
acrylamide formed during heating of different human foodstuffs. Acrylamide levels in foodstuffs were
analyzed by an improved gas chromatographic-mass spectrometric (GC-MS) method after bromi-
nation of acrylamide and by a new method for measurement of the underivatized acrylamide by liquid
chromatography-mass spectrometry (LC-MS), using the MS/MS mode. For both methods the
reproducibility, given as coefficient of variation, was ∼5%, and the recovery close to 100%. For the
GC-MS method the achieved detection level of acrylamide was 5 µg/kg and for the LC-MS/MS method,
10 µg/kg. The analytic values obtained with the LC-MS/MS method were 0.99 (0.95-1.04; 95%
confidence interval) of the GC-MS values. The LC-MS/MS method is simpler and preferable for most
routine analyses. Taken together, the various analytic data should be considered as proof of the
identity of acrylamide. Studies with laboratory-heated foods revealed a temperature dependence of
acrylamide formation. Moderate levels of acrylamide (5-50 µg/kg) were measured in heated protein-
rich foods and higher contents (150-4000 µg/kg) in carbohydrate-rich foods, such as potato, beetroot,
and also certain heated commercial potato products and crispbread. Acrylamide could not be detected
in unheated control or boiled foods (<5 µg/kg). Consumption habits indicate that the acrylamide levels
in the studied heated foods could lead to a daily intake of a few tens of micrograms.

KEYWORDS: Acrylamide; analysis; mass spectrometry; cooking; food; carcinogen

INTRODUCTION for further studies to provide incontestable proof that the origin
of the Hb adduct is acrylamide and, if so, to determine
In studies aimed at the identification of the causes of acrylamide sources and mechanisms of formation, as well as
background carcinogenesis, acrylamide has emerged as a factor an evaluation of the associated cancer risk.
that could be associated with a considerable cancer risk (1).
The occurrence of acrylamide in tobacco smoke (7), which
Studies of hemoglobin (Hb) adducts by mass spectrometric (MS)
could be observed in smokers as an increased level of the
methods have revealed background exposures to many reactive,
corresponding Hb adduct (5), indicated that acrylamide is formed
probably mutagenic and carcinogenic, compounds in humans
during incomplete combustion or heating of organic matter.
(2-4). For instance, a “background signal”, which corresponds
Furthermore, lower background levels of this Hb adduct were
to the Hb adduct to N-terminal valine from acrylamide, N-(2-
observed in wild animals when compared to humans and
carbamoylethyl)valine, has been regularly observed in unex-
laboratory animals (to be published), thought to be due to intake
posed control persons (5, 6). In connection with studies of
of unheated food in wild animals. A hypothesis that acrylamide
occupational exposure to acrylamide this background level was
is formed in cooking was confirmed in animal experiments by
actualized (1, 6). The background levels observed in Swedish
verification of the identity of the acrylamide adduct in Hb by
adults indicate an average daily intake of acrylamide that
comprehensive MS/MS analysis and the demonstration that the
approaches 100 µg, which could correspond to a non-negligible
cancer risk to the general population (1). This observation called increased adduct levels were compatible with expectation from
the contents of acrylamide determined in fried feed (8). The
present paper concerns the analysis of acrylamide in certain
* Author to whom correspondence should be addressed (telephone +46- human foodstuffs heated in cooking or manufacturing and the
8-162000; fax +46-8-152561; e-mail margareta.tornqvist@mk.su.se).
† Stockholm University. development of improved methodology for the analysis of
‡ AnalyCen Nordic AB. acrylamide in foodstuffs.
10.1021/jf020302f CCC: $22.00 © 2002 American Chemical Society
Published on Web 07/17/2002
Acrylamide in Cooked Foods
EXPERIMENTAL PROCEDURES
Chemicals. (13C3)Acrylamide (99%) was obtained from CIL
(Andover, MA); acrylamide (99+%) and N,N-dimethylacrylamide were
obtained from Sigma-Aldrich (Stockholm, Sweden). Bromine (g99.5%,
p.a.), hydrobromic acid (48%, p.a.), potassium bromide (g99%), sodium
thiosulfate pentahydrate (g99.5%), and sodium sulfate anhydrous
(g99%) were obtained from Merck (Darmstadt, Germany). All other
solvents and chemicals used for the analysis of acrylamide were of
analytical grade.
CAUTION: Acrylamide, (13C3)acrylamide, and N,N-dimethylacryl-
amide are hazardous and must be handled carefully.
Analytical Instruments. The quantification of acrylamide in food
was performed on a Hewlett-Packard (HP) 5890 gas chromatograph
(GC) coupled to an HP 5989A quadrupole mass spectrometer (MS).
The routine GC column was a BPX-10 fused silica capillary column
(30 m × 0.25 mm i.d., 0.25 µm film thickness; SGE, Ringwood,
Australia); similar columns were also used. Confirmatory analyses of
acrylamide were also made by liquid chromatography-tandem mass
spectrometry (LC-MS/MS) with electrospray positive ionization (ESI+)
using a Micromass Quattro Ultima coupled to an Agilent 1100 HPLC
(1100 binary pump, 1100 micro vacuum degasser, 1100 thermostated
autosampler, 1100 thermostated column compartment) with a Hypercarb
column (50 × 2.1 mm, 5 µm; ThermoHypersil) together with or without
a precolumn Hypercarb Guard (10 × 2 mm; ThermoHypersil).
Analysis by GC-MS of Acrylamide in Foodstuffs. The analysis
of acrylamide by GC-MS was performed with a simplified version of
the method previously applied for the analysis of laboratory animal
feed as the sample matrix (8), using brominated acrylamide as the
analyte (9, 10).
The preparation of samples for analysis involved mixing (with a
Waring 700 G blender) of 10 g of sample with water (100 mL),
followed by filtration (through a Sartorius glass-fiber filter; article no.
13400-90-S; Göttingen, Germany) and purification of the filtrate on a
graphitized carbon black column (Carbograph 4; 7 mm × 12 mm i.d,
1 g of carbon; LARA S.r.l., Rome, Italy). The internal standard, N,N-
dimethylacrylamide (500 µL from a stock solution in water), was added
(see also comment on new internal standard below).
The samples were derivatized through bromination by using potas-
sium bromide (7.5 g), hydrobromic acid (acidification to pH 1-3), and
saturated bromine water (10 mL) according to the method of Castle et
al. (9, 10). The sample was kept at 4 °C overnight, and the excess
bromine was decomposed by adding sodium thiosulfate (1 M) as drops
until the yellow color disappeared. Sodium sulfate (15 g) was added,
and the solution was exctracted with ethyl acetate/hexane [2 × 20 mL,
1:4 (v/v)]. The two pooled organic fractions were dried and evaporated
with a rotary evaporator to ∼200 µL.
The modifications in sample preparation in the present study
compared to the earlier method (8) are that final purification by filtration
and gel permeation chromatography (GPC) has been excluded. Fur-
thermore, during the course of the work (after initial studies on
hamburger meat) the internal standard, N,N-dimethylacrylamide, was
replaced with (13C3)acrylamide, to improve the accuracy and reproduc-
ibility of measurements. The new internal standard (1 mL of a stock
solution of 1 µg/mL of water) was added to the samples, at the initial
step, simultaneously with the water, in the workup procedure.
Two microliters of the samples was injected splitless, injector
temperature of 250 °C, on the GC-MS for analysis. The temperature
program for the GC was as follows: isothermal for 1 min at 65 °C,
increased at a rate of 15 °C/min to 250 °C, and isothermal for 10 min.
The analysis was performed using electron ionization (70 eV) and
selected ion monitoring. The ions monitored for identification of the
analyte, 2,3-dibromopropionamide, were [C3H581BrNO]+ ) 152 (100%),
[C3H579BrNO]+ ) 150 (100%), and [C2H379Br]+ ) 106 (65-70%) using
m/z 150 for quantification. The ions monitored for identification of
the internal standard, brominated to 2,3-dibromo(13C3)propionamide,
were [13C3H581BrNO]+ ) 155 (100%) and [13C2H381Br]+ ) 110 (65-
70%) using m/z 155 for quantification (a variation of (10% in ratio
between fragment ions is allowed for identification). The ions monitored
for identification and quantification when using N,N-dimethylacryla-
J. Agric. Food Chem., Vol. 50, No. 17, 2002 4999
mide, brominated to 2,3-dibromo-N,N-dimethylpropionamide, as an
internal standard were [C5H981BrNO]+ ) 180 and [C5H979BrNO]+ )
178.
Quantification was performed by comparison with a calibration curve
(0.5-50 µg/L water, corresponding to 5-500 µg/kg). Samples with
concentrations >500 µg/kg of acrylamide were diluted up to a factor
of 6 in the first step when food was mixed with water. Recovery tests
were repeatedly performed by quantification of acrylamide in different
(both raw and heated) foodstuffs before and after the addition of
acrylamide.
Analysis by LC-MS/MS of Acrylamide in Foodstuffs. The samples
were homogenized, and 100 mL of water and 1 mL of the internal
standard, (13C3)acrylamide (1 µg/mL of water), were added to 10.0 g
of the homogenized sample. The samples were centrifuged in 12 mL
Pyrex glass tubes, and the particle-free supernatant was further
centrifuged (at 14000 rpm for 10 min) in two Eppendorf tubes (1.5
mL/tube). An Isolute Multi-Mode SPE column (300 mg; International
Sorbent Technology Ltd.), activated with acetonitrile (1 mL) and washed
with water (2 + 2 mL), was used to trap nonpolar interferences by
adsorption in the recombined supernatant (3 mL). The first milliliter
of the filtrate was discarded and the rest passed through a syringe filter
(0.45 µm; Sartorius minisart hydrophilic, article no. 17598). Five
hundred microliters was ultrafiltered (Microcon YM-3, article no.
42404, Millipore) in an Eppendorf centrifuge (14000 rpm, 10 min) until
200 µL had passed through.
The samples were analyzed by LC-MS/MS (ESI+) (at ambient
temperature) using the column given above, using deionized water as
mobile phase, with a flow rate of 0.2 mL/min for 6.1 min (analytes
recorded), washing with 80% aqueous acetonitrile (4 min at a flow
rate of 0.4 mL/min), followed by reconditioning with water (0.2 mL/
min, 10 min) between sample injections (20 µL). The electrospray
source had the following settings (with nitrogen): capillary voltage,
3.2 kV; cone voltage, 50 V; source temperature, 125 °C; desolvation
temperature, 350 °C; cone gas flow, 211 L/h; and desolvation gas flow,
653 L/h. Argon (2.5 mbar) was used as collision gas.
Acrylamide was identified by multiple reaction monitoring (MRM).
The precursor ion [M + H]+ ) 72 was fragmented, and product ions
[H2CdCHsCdNH]+ ) 54 (collision energy ) 16 eV) and [H2Cd
CHsCdO]+ ) 55 (collision energy ) 11 eV) were monitored (ratio
between product ions 1:35 ( 20%). The ion m/z 55 was used for
quantification. Monitored product ion for the internal standard was
[13C3H3O]+ ) 58 from precursor ion [M + H]+ ) 75. The collision
energy was 11 eV. For all MRM transitions the dwell time was 0.3 s
with the cone voltage given above. Product ion spectra of acrylamide
and analyte were compared at collision energies of 10 and 20 eV,
respectively.
Quantification was performed through comparison with a calibration
curve with five concentrations (1-500 µg/L water, corresponding to
10-5000 µg/kg). The repeatability, recovery, and reproducibility of
the LC-MS/MS method were evaluated, and the method was compared
with the GC-MS method.
Laboratory Cooking. Heating under laboratory conditions by
frying-searing was carried out in a thermostated iron frypan (11). The
temperature in the frypan and in the browning zone of food during
heating was controlled with a CIE model 307 digital thermometer,
connected with a probe (accuracy (2.2 °C; Clas Ohlson, Stockholm,
Sweden). Heating in a microwave oven was performed with a Whirlpool
model MD 121, equipped with a rotating plate, at 750 W (nominal
power). Heating in the oven during controlled conditions was performed
in a temperature-programmed GC oven (Hewlett-Packard model
5790A). The homogenization of potato was performed with a Philips
chopper, model HR 1392.
All foodstuffs used in the cooking experiments were obtained from
grocery stores. In the experiments minced or grated foodstuff formed
into 9 cm diameter patties of ∼60 g portions was used, unless otherwise
stated. After cooking, the samples were immediately frozen (-20 °C)
prior to analysis. The samples were sent from Stockholm to Lidköping
in dry ice for analysis of acrylamide content.
Influence of Temperature and Time on Acrylamide Formation
during Heating of Beef by Frying. Patties of beef (commercially
available as frozen hamburgers) were fried (without addition of oil) at
5000 J. Agric. Food Chem., Vol. 50, No. 17, 2002 Tareke et al.

temperatures of 160, 180, 200, 220, and 240 °C for 3 min on each
side. Two raw patties were used as controls, and two samples were
prepared at each of the temperatures 160, 200, and 240 °C. In addition,
the influence of heating time was studied by heating single samples of
lean minced beef for 1 or 4 min on each side at 200 °C. In the first
experiments the temperature was also controlled with a digital
thermometer, connected with a probe.
Comparison of Different Foodstuffs with Regard to Acrylamide
Formation during Heating by Frying. In one experiment the influence
of protein source on the formation of acrylamide was studied in protein-
rich foods. Duplicate samples of fish fillet (cod), lean beef, lean pork
(one sample lost), chicken fillet, and soy flour (soaked in water) were
heated. In another experiment grated potato (n ) 2), grated beetroot
(n ) 2), and spinach (n ) 1) were studied as examples of carbohydrate-
rich foods. In both experiments the foods were heated in a frypan for
2.5 min on each side at 220 °C, without addition of oil.
Influence of Cooking Method on Acrylamide Formation. The
influence of the cooking method with regard to acrylamide formation
was studied by comparing heating in a frypan, boiling, and microwave
heating. Lean beef (n ) 3), minced fish (cod) (n ) 1), grated potatoes
(n ) 3), and boiled potato mashed (n ) 2) were heated for 2.5 min on
each side in a frypan at 220 °C (without addition of oil). Patties of
minced fish (cod) (n ) 2), lean minced beef (n ) 4), and grated potatoes
(n ) 4) were boiled (for up to 20 min). Broth from boiling of lean
beef, cod, and potatoes was sampled for analysis (n ) 3 + 2 + 4).
Minced fish (cod) (n ) 2) and grated potato (n ) 2) were heated in a
microwave oven for 3 min on both sides at 750 W (nominally). Raw Figure 1. Analysis of acrylamide content in laboratory-boiled minced beef
potato was analyzed as control sample. (Samples of raw beef were prior to and after addition of acrylamide and of potato crisps with high
analyzed in the first experiment, described above.) content of acrylamide; GC-MS (EI) chromatograms of four selected ions
Influence of Time on Acrylamide Formation at Microwave from the analyte of acrylamide (2,3-dibromopropionamide) and from the
Heating of Potato. The influence of the duration of microwave heating
internal standard [(13C3)acrylamide] after bromination of the samples
on acrylamide formation in potato was studied. Potato (∼150 g) was
peeled, homogenized, and subdivided into smaller 30.0 g portions in (variation in retention times is due to use of different GC columns, PAS
Erlenmeyer flasks (250 mL). The samples were heated in a microwave 1701 and BPX-10, respectively). In the figure m/z 152 is omitted because
oven (750 W nominally) for 50, 100, and 150 s, followed by weighing of contribution from the internal standard at low levels.
for determination of weight loss, mainly water. As a control a
nonheated, homogenized portion was used. (8) was further simplified by excluding the GPC cleanup of
Influence of Temperature on Acrylamide Formation in Oven- the extract, which mainly had the aim of removing fat from the
Heated Potato. Commercially available French-fried potatoes, prepared sample and preventing deposits in the ion source. The reproduc-
for heating in home ovens (recommended heating time at 225 °C ) ibility was improved by changing the internal standard to 13C-
25-30 min), were heated in a temperature-programmed GC oven. Fries substituted acrylamide and by addition of the internal standard
of about equal size and length were selected to make 50.0 g samples.
earlier in the workup procedure.
The samples of the frozen French fries were then heated in a glass
bowl using the following temperature program for the oven: 60 °C The calibration curve was linear in the range studied (5-
for 1 min and then 50 °C/min until the final temperature was reached 500 µg/kg). In recovery tests with addition of acrylamide (5-
(100-220 °C). The time period of maximal heating temperature was 1000 µg/kg) to different foods containing <5->2000 µg/kg
adjusted (19-14.2 min) to obtain about the same total heating time of “natural” acrylamide, the recovery with the original internal
(21 min). The samples were then cooled to 60 °C, weighed, and stored standard was 98% (SD ) 26%) and with the new internal
at -20 °C until analysis. The French fries heated at 180 and 200 °C standard, 98% (SD ) 7.5%). The coefficient of variation (CV)
were browned as normally served. for the GC-MS method using the original internal standard is
Acrylamide Contents in Food Items from Restaurants, Etc. Potato ∼15% in the range of 5-100 µg/kg and above this range,
products and other foodstuffs and products were obtained from ∼10%. The reproducibility of the method was improved with
restaurants or from grocery stores, and samples were taken and analyzed
the new internal standard to ∼5% in the analytical range. The
(the same day as purchased) for comparison with acrylamide content
in the laboratory-prepared foodstuffs. Details of the prepurchase limit of detection was 5 µg/kg. Figure 1 illustrates analyses
processing were not available, and variations with respect to composi- with the improved GC-MS method of samples with high and
tion and cooking etc. method were evident. Also, for evaluation of low contents of acrylamide (using different GC columns).
analytical methods commercial foodstuffs were used for analysis. An LC-MS/MS method was developed to achieve confirma-
tion by an independent method not involving derivatization of
RESULTS AND DISCUSSION acrylamide (Figure 2). The calibration curve was linear in the
Methods for Analysis of Acrylamide. The analyses of range of 10-5000 µg/kg (r2 > 0.9999). The repeatability,
acrylamide were first performed by GC-MS following bromi- recovery, and reproducibility obtained with the method are
nation, and later the contents of acrylamide were confirmed, shown in Tables 1-3, respectively. The repeatability (Table
without derivatization, by LC-MS/MS. The methods using 1) of measurements was determined on samples with acrylamide
bromination of acrylamide (9) and GC-MS analysis are similar contents in the range of 40-1700 µg/kg, with CV of ∼5% (2-
to the U.S. EPA method for analysis of acrylamide in water 9%). The recovery (Table 2) was determined as 99.5% (SD )
(10). This method has been used in our laboratory for analysis 6.5%). The reproducibility (Table 3) was determined as CV to
of acrylamide in water (12) and, after further development, for be in the order of 5% (2-9%). The detection level was ∼10
analysis of laboratory animal feed as sample matrix (8). In the µg/kg for the samples in the present study but may vary
present work the modified GC-MS method used for animal feed depending on the matrix. The LC-MS/MS method was com-
Acrylamide in Cooked Foods J. Agric. Food Chem., Vol. 50, No. 17, 2002 5001

Figure 2. LC-MS/MS analysis of acrylamide in French-fried potatoes (a) by monitoring the precursor ion m/z 72 and product ions m/z 55 and 54,
quantified with the internal standard precursor ion m/z 75 and product ion m/z 58. (b) Analysis of a blank sample.

Table 1. Repeatability Test of the LC-MS/MS Methoda Table 2. Examples of Recovery Tests of LC-MS/MS Method (Samples
IV and V, Replicate Analysis)a
mean CV
sample acrylamide (µg/kg) (µg/kg) (%) before added after
I bread 35, 33, 41, 39, 39 37 8.9 addition amount addition recovery
II bread 46, 49, 51, 51, 47 49 4.7 sample (µg/kg) (µg/kg) (µg/kg) (%)
III French fries 446, 451, 416, 421, 388 424 6.0 I rye 3.3 10 13 97
IV potato crisp 1584, 1523, 1530, 1362, 1479 1496 5.6 II oat 7.3 10 17 97
V bread 1704, 1715, 1741, 1765 1731 1.6 III bread 37 50 93 112
IV French fries 424 600 1036, 970 102, 91
a Each food sample was split into five aliquots prior to homogenization, workup, V potato crisps 1496 1500 2978, 2899 99, 94
and analyses during one day and with the same laboratory staff. VI potato crisps 1678 1000 2716 104

a Percent recovery of amount acrylamide added to samples with analyzed

pared with the GC-MS method with (13C3)acrylamide as internal
standard (Table 4). By orthogonal regression of log-transformed
values, assuming the same CV, LC-MS/MS values were found
to be 0.99 [0.95-1.04; 95% confidence interval (CI)] of the
GC-MS values. The LC-MS/MS method is simpler than the GC-
MS method and will be the method of choice for routine analysis
of acrylamide. Both methods have been accredited (according
to ISO 17025; 13). It may be mentioned that an LC-MS/MS
acrylamide contents.
method for analysis of acrylamide in foodstuffs has also been
developed by the Swedish National Food Administration (NFA)
(34).
CV for duplicate samples in the complete study was up to
20% (with respect to, e.g., frying and sample workup, with
analysis on the same occasion by the same method). A high
5002 J. Agric. Food Chem., Vol. 50, No. 17, 2002 Tareke et al.

Table 3. Reproducibility Test of the LC-MS/MS Methoda Table 5. Retention Times Obtained for 2,3-Dibromopropionamide and
Acrylamide with Different Columns by the GC-MS and LC-MS/MS
analysis I analysis II analysis III Methods, Respectivelya
sample (µg/kg) (µg/kg) (µg/kg) CV (%)
I bread 37 41 44 8.6 length i.d. film thickness retention
II French fries 424 434 1.6 GC columnb (m) (mm) (µm) timec (min)
III potato crisps 1496 1544 2.2 SE30 25 0.32 0.5 7.3
IV crispbread 1731 1564 7.2 PAS1701 25 0.32 0.25 9.5
DB-1701P 30 0.32 0.25 10.0
a An aliquot of the food samples was taken out on different days, homogenized, BPX-10d 30 0.25 0.25 11.2
and analyzed by different laboratory staff, on different occasions.
length i.d. chromatographic retention
Table 4. Comparison Test of the LC-MS/MS Method versus the LC column (mm) (mm) conditions timee (min)
GC-MS Method, Both Methods Using (13C3)Acrylamide as Internal Shodex Rspak 150 4.6 eluent, acetic acid in 4.0
Standard DE-413 water at pH 2.6;
flow ) 1.0 mL/min
GC-MS LC-MS/MS difference Hypercarbd 50 2.1 eluent, water (without 3.5
sample (µg/kg) (µg/kg) (%) buffer);
I hamburger 14 14 0 flow ) 0.2 mL/min
II hamburger 23 23 0
III French fries 661 684 3.5 a Stated retention times are approximate values. In all cases the analyte coeluted
IV potato crisps 1538 1544 0.39 with the chosen internal standard. b The used chromatographic conditions were
V potato crisps 1800 1678 −6.8
equivalent for the different columns and the same as described under Experimental
Procedures. c The retention time of the analyte, 2,3-dibromopropionamide, and that
of the brominated internal standard, (13C3)2,3-dibromopropionamide, were identical.
stability of acrylamide in stored samples was indicated in the d Column used for measurements in the present study. e The retention time of
repeated analysis of fried potato stored for 1 year in the freezer. acrylamide and that of the internal standard, (13C3)acrylamide, were identical.
Identification of Acrylamide. The identity of acrylamide is Retention time for column without precolumn given.
supported as follows: In the various foodstuff matrices acryl-
amide was determined by two different procedures for workup, levels (micrograms per kilogram of heated product) in fried
chromatography, and detection as described under Experimental hamburgers from the initial experiment with frying of hamburger
Procedures. Whereas the procedure for GC-MS analysis involves meat at different temperatures. A significant dependence of
bromination at low pH and analysis of brominated samples at acrylamide formation on temperature was demonstrated. In raw
high temperatures, the procedure for analysis by LC-MS/MS is hamburgers the acrylamide content was below the detection level
more lenient with respect to acidity and temperature. In the (5 µg/kg). A dependence of content of acrylamide on frying
procedure for GC-MS analysis acrylamide is derivatized to 2,3- time was also indicated (data not shown).
dibromopropionamide by bromination of the ethylenic double In subsequent experiments different types of foodstuffs heated
bond according to well-known procedures (9, 10). Analysis by via cooking methods were studied with respect to the formation
LC-MS/MS involves direct determination of underivatized of acrylamide in the heated foodstuffs. The level of acrylamide
acrylamide. In both methods (13C3)acrylamide was used as was below the detection level (5 µg/kg) in boiled or raw beef
internal standard and measured as brominated derivative or and potato, in boiled fish, as well as in broth from the boiling
without derivatization, respectively. When applied to the same tests. Following controlled heating, protein-rich foods exhibited
samples the two methods despite the differences in the acrylamide concentrations between 5 and 50 µg/kg, with lower
methodology gave concordant results (cf. Table 4). levels in fish. In the experiments with heating of carbohydrate-
The analysis of acrylamide was performed on four different rich foods, relatively higher contents of acrylamide, 150-1000
GC columns and two different LC columns (Table 5). In all µg/kg, were measured. Following microwave-heating of fish
separations the analytes exhibited the same retention times as and potato, no detectable levels of acrylamide were found in
the corresponding internal and external standards. fish. Detailed data are given in Table 6 and illustrated in Figure
Analysis by an MS/MS method (monitoring of product ions 5.
of a precursor ion, MRM) used for the underivatized acrylamide These laboratory experiments were followed by studies of
in this study gives stronger evidence of identity than MS analysis acrylamide contents in commercially prepared foodstuffs pur-
only. The acrylamide content in potato chips was verified by chased from restaurants or grocery stores. Analysis of the
recording product ion spectra in LC-MS/MS analysis. The selected foods, mainly potato products, gave results compatible
spectra are identical for the standard and the analyte, at 10 and with those obtained after cooking under laboratory conditions
20 eV, respectively (Figure 3). This is a strong support for the (Figure 5; Table 6). French-fried potatoes and potato crisps
identity of the analyte. In addition, further support for the exhibited relatively high levels of acrylamide [median values
identity is that several ions are monitored for the analyte in of 424 µg/kg (n ) 5) and 1739 µg/kg (n ) 6), respectively].
both the GC-MS analysis and the LC-MS/MS analysis and that Large variations of acrylamide levels in similar foods were
the relative ion abundances are the same for the analyte as for observed and could also be expected when prepared com-
the standard. With the aid of the 13C-substituted internal standard mercially, because of differences in heating time, etc. This was
the product ion spectrum in the LC-MS/MS analysis as well as observed, particularly for French fries and potato crisps. There
the mass spectrum in GC/MS was interpreted (see Experimental are also variations between samples of fried potatoes. It should
Procedures). be noted that the commercially fried potato pancake contained
Studies on alternative techniques for identification will be not only grated potato (as in the laboratory frying) but also is
pursued. mixed with egg, milk, and flour, which might have had an
Quantities of Acrylamide Formed during Cooking of influence on acrylamide formation during heating. Boiling of
Different Foodstuffs. Figure 4 shows measured acrylamide potatoes prior to frying seemed to reduce the formation of
Acrylamide in Cooked Foods
Figure 3. LC-MS/MS (ESI+); comparison between product ion spectra from precursor ion m/z 72 obtained from an acrylamide standard (10 µg/mL) and
an analyte (in potato chips) recorded at 10 and 20 eV collision energy and scan range of 20−80 amu.
Figure 4. Acrylamide concentrations (micrograms per kilogram) in
hamburgers (single and duplicate samples) dependent on frypan tem-
peratures (3 min of frying time on each side).
acrylamide (tested in parallel experiments), an observation that
will be addressed in future experiments.
Other carbohydrate-rich foods, such as crispbread (Swedish
knaeckebroed), showed variable acrylamide levels (see Table
6, e.g., in differently prepared products of crispbread). Toasting
of soft bread produced a moderate formation of acrylamide
(increasing from 13 to 53 µg/kg; one trial). Acrylamide in
crispbread, which is usually stored at room temperature for
longer periods of time, also demonstrated that acrylamide is
relatively stable in foods.
The influence of heating time on the formation of acrylamide
in potatoes during heating in a microwave oven was followed
in a laboratory experiment, with simultaneous measurement of
weight loss (Table 7). When homogenized raw potato was
heated for 50, 100, and 150 s, the highest levels of acrylamide
were present after heating for 150 s (mean ) 4400 µg/kg of
heated product). A heating time of 100 s resulted in ∼100 times
lower levels. Measurement of weight loss showed that the
J. Agric. Food Chem., Vol. 50, No. 17, 2002 5003
samples heated for 100 s still contained water (∼40%), whereas
the samples after 150 s in principle had lost all water. There is
a high water content in potato (∼79%) (14), and it appears that
water retention in the potato protected against pyrolysis. Visually
there are obvious differences between the samples; after 100 s,
only minor parts were browned, whereas the samples cooked
for 150 s were thoroughly browned with totally black areas
present (not edible).
The influence of temperature on the formation of acrylamide
in potato was also studied with French-fried potato heated at
controlled conditions in an oven. The relatively high amount
of acrylamide in the control sample (146 µg/kg) is due to heat
treatment in the production (seen as weak browning). Laboratory
heating for 19 min at 100 °C had no effect on the acrylamide
content (see Table 8 and Figure 6). A small increase (∼ +30
µg/kg) in the acrylamide level measured at 120 °C indicates
that the temperature needed for formation of acrylamide is >100
°C.
Aspects of Formation of Acrylamide during Heating of
Foodstuffs. The hypothesis that acrylamide could be formed
during the heating of food was earlier tested in animal
experiments and supported through analysis of fried laboratory
animal feed (8). A second hypothesis was that protein was
involved in the formation of acrylamide, a hypothesis that
seemed to be supported in our initial analyses of acrylamide
levels in fried hamburger meat within the present study (Figure
4). However, the content of acrylamide in fried beef and other
protein-rich foods was low compared to that in the fried
laboratory animal feed (rich in carbohydrates) as previously
studied (8). Therefore, also other types of foodstuff were tested,
for example, potato. Analyses of fried potato revealed acryl-
amide contents at 10-100 times higher levels than in the
protein-rich foods. Frying of another type of carbohydrate-rich
foodstuff (e.g., beetroot, Figure 5) gave similar results.
5004 J. Agric. Food Chem., Vol. 50, No. 17, 2002 Tareke et al.

Table 6. Levels of Acrylamide Analyzed in Single Samples of Different increasing temperature (Figures 4 and 6). Compared with
Foodstuffs (Micrograms per Kilogram of Heated Foodstuff)a HCAs, acrylamide (with highest contents observed in carbo-
hydrate-rich foods), does not contain any amino acid residue
Laboratory-Prepared Foods and is certainly formed by other pathways, possibly via reactive
fried food A-1 A-2 A-5 median three-carbon units formed in the heating of carbohydrates, for
beef, minced 20; 22 15; 15; 17 17 example, monosaccharides (18, 19), with, for example, am-
chicken, minced 16; 41 28 monia, as nitrogen source. The formation of acrylamide shows
cod, minced <5; <5 11 similarities with the Maillard reaction.
pork, minced 52
soymeal beef 15; 16 16 Estimated Dietary Intake of Acrylamide. It was prelimi-
potato, grated 730; 780 447 394; 310 447 nary estimated that acrylamide in four of the investigated
potato, boiled, mashed 201; 144 172 products together could lead to a daily intake of a few tens of
beetroot, grated 810; 890 850 micrograms of acrylamide. This estimation was also verified
spinach, grated 112
by using Swedish consumption statistics (20) (K. Svensson,
NFA, Uppsala, Sweden, personal communication). The average
microwave-heated food A-2 median
level of the background Hb adduct level from acrylamide
cod <5; <5 observed in earlier studies of humans with no known exposure
potato, grated 455; 650 551
was preliminarily estimated to result from a daily intake of ∼100
µg of acrylamide by an adult person (1). It has to be recalled
boiled food A-2 A-4 A-5 median
that these calculations of intake from the average level of
potato (boiled or rawb) <5; <5 <5; <5 <10*b background Hb adduct from acrylamide are based on parameter
potato, broth <5; <5 <5; <5
beef (boiled or rawb) <5;b <5b <5; <5 <5; <5
values in the rather complex pharmacokinetics of acrylamide
beef, broth <5 <5; <5 (21); these values require further studies. Also, it cannot be
cod <5; <5 excluded that endogenous production of acrylamide gives a
cod, broth <5; <5 certain contribution to the background level of the Hb adduct.
Estimated Health Risks Associated with Intake of Acryl-
Restaurant-Prepared/Purchased Foods
amide. The present study is part of a series of investigations
food A-3 A-4 A-5 medianc with the main purpose of clarifying the role in background
hamburger 14/14*; 23/23* 18 carcinogenesis of identifiable chemical mutagens of exogenous
pork (fried) 45 or endogenous origin.
“falu” sausage (fried) <5
French fries 732 314; 327 661/684*; 424* 424 Disease-epidemiological investigations have been unable to
potato pancake 167 either confirm or disprove human carcinogenicity of acrylamide,
potato cubes 313 even in occupationally exposed cohorts (22, 23). Thus, human
potato crisps 1300; 1538/1544*; 1739 data useful for cancer risk estimation are not available. The
1800/1678* 2148*; 3897*;
magnitude of the cancer risk from acrylamide has therefore been
1496*
crispbread, 208 37*; 1731* 208 assessed from animal experiments (two-year cancer tests with
three types rats) (24, 25), using different linear no-threshold models.
bread, white 13;d 53d 49* According to WHO, a lifelong (70 years) intake of 1 µg of
rye <5* acrylamide per day would be associated with a lifetime cancer
oat 7.3*
beer (dark and lager, <5; <5; <5
risk of 1 × 10-5 (26). This value is derived by using a linearized
three types) multistage model without efforts to allow for differences in
metabolism between species. A scaling via dose per unit of body
a A-1−A-5 denote different analytical series. Analysis with GC-MS; analysis with surface area according to the U.S. EPA (27) leads to an ∼6
LC-MS/MS marked with an asterisk (*). Acrylamide concentrations determined in times higher lifetime risk, 6 × 10-5 per µg × day.
the samples by both GC-MS and LC-MS/MS have been separated by a slash (/). The mutagenic and carcinogenic factor in acrylamide expo-
b Samples before boiling. Samples of raw beef from the first experiment (Figure
sure is assumed to be the epoxy metabolite, glycidamide (28-
4). c For repeated analysis of the same sample, separated by / mean values are 30). The cancer risk has also been estimated on the basis of the
used. d Before and after toasting. dose of glycidamide, with approximately the same result as
From the results on analyses of foodstuffs heated under obtained with the U.S. EPA procedure (1). With a multiplicative
laboratory conditions and from the analyses of commercially risk model, shown to be adaptable to the acrylamide cancer test
available foodstuffs it can be concluded that of the studied data (31) based on the dose of this epoxide, the value for lifetime
cooking conditions, except boiling, all lead to pronounced cancer risk is somewhat higher than the above value estimated
acrylamide formation in potato. according to the U.S. EPA (1).
In preliminary experiments aimed at finding cooking condi- The drinking water guideline of WHO for acrylamide is 0.5
tions that might reduce or prevent the formation of acrylamide, µg/L (26), corresponding to an intake of 1 µg/day at the
addition of oils, antioxidants, or argon atmosphere during the consumption of 2 L/day. In 2003 the recommended limit will
frying of beef was tested. These measures had minor or become 5 times lower in the European Union (32).
nonsignificant acrylamide-reducing effects (data not shown). Higher daily intakes of acrylamide may lead to neurotoxic
Furthermore, it should be noted that in the present study symptoms, which, in contrast to tumors, exhibit a nonlinear
evaporated acrylamide was not measured (boiling point esti- dose-response. A no-effect threshold for light symptoms
mated to be ∼225 °C). appears to be between 800 and 2700 µg/day corresponding to
The classical cooking mutagens/carcinogens, first reported adduct levels of 300-1000 pmol/g of Hb (1, 6).
by Sugimura and his colleagues (15), consist of a number of Observations on Acrylamide Contents in Heated Foods
heterocyclic amines (HCAs) (16, 17), which are characterized and as Hemoglobin Adducts in Consumers’ Blood. Data from
by participation of different amino acids and association with different studies may be used to validate the identity and levels
Acrylamide in Cooked Foods J. Agric. Food Chem., Vol. 50, No. 17, 2002 5005

Figure 5. Acrylamide concentrations (micrograms per kilogram) (median and range) in laboratory-heated and commercial foodstuffs (cf. Table 5).

Table 7. Acrylamide Concentrations and Weight Loss in

Microwave-Heated Mashed Potato, Heated at a Power of 750 W, for
the Heating Times Given
heating time (s) na wt loss (SE) (%) acrylamide (SE) (µg/kg)
control 3 0 <5
50 2 27 (±1.5) <5
100 3 61 (±0.48) 47 (±16)
150 3 77 (±2.3) 4400 (±1150)

a Number of experiments.

Table 8. Acrylamide Concentrations and Weight Loss in Commercially

Prepared and Uncooked French-Fried Potatoes Heated in a
Programmed GC Oven (Total Heating Time ) 21 min)a Figure 6. Acrylamide concentrations (micrograms per kilogram), corrected
acrylamide, for weight loss, in French-fried potatoes heated in a temperature-
max temp acrylamide time (min) at wt loss corrected values programmed oven.
(°C) (µg/kg) max temp (%) (µg/kg)
control 146 0 0 146 was measured as an increase of acrylamide adducts to Hb, the
100 172, 174 19.0 14.9 146, 148 incremental adduct level being shown to be compatible (con-
120 217 18.2 19.9 174 sidering, e.g., the reactivity of acrylamide) with the content of
140 376 17.4 29.8 264 acrylamide in the feed (8). The adduct structure was verified
160 808 16.6 41.9 469
180 1965 15.8 49.0 1003 by interpretation of product ions in comparison with corre-
200 3479 15.0 54.3 1591 sponding ions from one unsubstituted and two different isotope-
220 5051 14.2 55.0 2273 substituted standards (8). It could further be added that the
preliminarily estimated intake of acrylamide in this study is
a Measured values were also corrected for weight loss. within the range expected from the background level of the
of acrylamide in heated foods. Evidence for the identity of adduct from acrylamide in humans. It further strengthens the
acrylamide in foodstuffs in the analyses presented in this study proof of identity that the valine adduct, N-(2-carbamoyl-2-
is summarized above. Strong supporting evidence for the identity hydroxyethyl)valine, from the prime acrylamide metabolite,
was also obtained from the analysis of Hb adducts formed in glycidamide, is regularly found in the same Hb samples as the
vivo through Michael addition of acrylamide to the amino acrylamide adduct and in approximately the proportion to the
termini, valines, of Hb and detached from the protein as acrylamide adduct expected from acrylamide exposure (unpub-
derivatives of N-(2-carbamoylethyl)valine. First, a background lished data).
level of the same adduct as observed in humans with exposure The present work has shown that a large contribution to the
to acrylamide (occupationally and by smoking) is observed in background level of the Hb adduct from acrylamide in humans
unexposed humans. The same adduct has also been observed probably originates from acrylamide formation in cooking and
in studies of laboratory animals treated with acrylamide and in food preparation. Relatively high acrylamide concentrations in
animals with accidental exposure to acrylamide. In a previous certain foodstuffs were recently presented by the Swedish
study with rats fed fried laboratory feed, an intake of acrylamide National Food Administration (NFA), which further explored
5006 J. Agric. Food Chem., Vol. 50, No. 17, 2002
our present, initial observations; the NFA results are concordant
with our data (33).
This pilot study should be regarded as a first attempt to
determine acrylamide in frequently consumed foodstuffs and
also to demonstrate the importance of the cooking technique in
the formation of acrylamide. The study demonstrates the need
for further studies of the formation of acrylamide during cooking
of foods and of the presence of this chemical in foods consumed.
In future studies the determination of Hb adducts will be an
important tool for measurement of the dietary intake of
acrylamide.
ACKNOWLEDGMENT
We are most grateful to Marcus Sunnelöv, B.Sc., for skillful
assistance with LC-MS/MS analysis and Professors L. Ehren-
berg and A° . Bergman and Associate Professor Fredrik Granath
for fruitful discussions and critical viewpoints.
NOTE ADDED AFTER ASAP POSTING
The original posting made July 18, 2002, contained an error
in m/z value on page B and omitted reference numbers on p H.
These errors have been corrected in this posting.
LITERATURE CITED
(1) Törnqvist, M.; Bergmark, E.; Ehrenberg, L.; Granath, F. Risk
Assessment of Acrylamide; Report 7/98; Swedish Chemicals
Inspectorate: Solna, Sweden, 1998 (in Swedish).
(2) Törnqvist, M.; Kautiainen, A. Adducted proteins for identification
of endogenous electrophiles. EnViron. Health Perspect. 1993,
99, 39-44.
(3) Törnqvist, M. Ethylene oxide as a biological reactive intermediate
of endogenous origin. In Biological ReactiVe Intermediates V;
Snyder, R., et al., Eds.; Plenum Press: New York, 1996; pp 275-
283.
(4) Richter, E.; Branner, B. Biomonitoring of exposure to aromatic
amines: Haemoglobin adducts in humans. J. Chromatogr. B
2002, in press.
(5) Bergmark, E. Hemoglobin adducts of acrylamide and acrylonitrile
in laboratory workers, smokers, and nonsmokers. Chem. Res.
Toxicol. 1997, 10, 78-84.
(6) Hagmar, L.; Törnqvist, M.; Nordander, C.; Rosén, I.; Bruze, M.;
Kautiainen, A.; Magnusson, A. L.; Malmberg, B.; Aprea, P.;
Granath, F.; Axmon, A. Health effects of occupational exposure
to acrylamide using hemoglobin adducts as biomarkers of internal
dose. Scand. J. EnViron. Health 2001, 27, 219-226.
(7) Schumacher, J. N.; Green, C. R.; Best, F. W.; Newell, M. P.
Smoke composition. An extensive investigation of the water-
soluble portion of cigarette smoke. J. Agric. Food Chem. 1977,
25, 310-320.
(8) Tareke, E.; Rydberg, P.; Karlsson, P.; Eriksson, S.; Törnqvist,
M. Acrylamide: A cooking carcinogen? Chem. Res. Toxicol.
2000, 13, 517-522.
(9) Castle, L.; Campos, M. J.; Gilbert, J. Determination of acrylamide
monomer in hydroponically grown tomato fruits by capillary gas
chromatography mass spectrometry. J. Sci. Food Agric. 1993,
54, 549-555.
(10) U.S. EPA. SW. 846, Test methods for eValuating solid waste,
Acrylamide by gas chromatography, Method 8032A; U.S.
Environmental Protection Agency: Washington, DC, 1996.
(11) Övervik, E.; Nilsson, L.; Fredholm, L.; Levin, Ö.; Nord, C. E.;
Gustafsson, J. A° . High mutagenic activity formed in pan-broiled
pork. Mutat. Res. 1984, 135, 149-157.
(12) SWEDAC. Accreditation SWEDAC Dnr. 97-4101-51.1125;
Swedish Board for Accreditation and Conformity Assessment
1998-03-02; Stockholm, Sweden.
(13) SWEDAC. Accreditation SWEDAC Dnr. 01-4262-51.1125;
Swedish Board for Accreditation and Conformity Assessment
2002-05-15; Stockholm, Sweden.
(14) U.S. Department of Agriculture (USDA). http://
www.nal.usda.gov.
Tareke et al.
(15) Sugimura, T.; Kawachi, T.; Nagao, M.; Yahagi, T.; Seino, Y.;
Okamoto, T.; Shudo, K.; Iitaka, Y.; Itai, A. Mutagenic principles
in tryptophan and phenylalanine pyrolysis products. Proc. Jpn.
Acad. 1977, 53, 58-61.
(16) Skog, I. K.; Johansson, M. A. E.; Jägerstad, I. M. Carcinogenic
heterocyclic amines in model systems and cooked foods: A
review on formation, occurrence and intake. Food Chem. Toxicol.
1998, 36, 879-896.
(17) Sugimura, T. Nutrition and dietary carcinogens. Carcinogenesis
2000, 21, 387-395.
(18) Weenen, H. Reactive intermediates and carbohydrate fragmenta-
tion in Maillard chemistry. Food Chem. 1998, 62, 393-401.
(19) Kabyemela, B. M.; Adschiri, T.; Malaluan, R. M.; Arai, K.
Glucose and fructose decomposition in subcritical and super-
critical water: Detailed reaction pathway, mechanisms, and
kinetics. Ind. Eng. Res. 1999, 38, 2888-2895.
(20) Swedish dietary survey carried out in 1997-1998 (Riksmaten).
Vaar Foeda 1999, 53 (1), 24-27.
(21) Calleman, C. J. The metabolism and pharmacokinetics of
acrylamide: Implications for mechanisms of toxicity and human
risk. Drug Metab. ReV. 1996, 28 (4), 527-590.
(22) Marsh, G. M.; Lorraine, J. L.; Youk, A. O.; Schall, L. C.
Mortality patterns among workers exposed to acrylamide: 1994
follow up. Occup. EnViron. Med. 1999, 56, 181-190.
(23) Granath, F.; Ehrenberg, L.; Paulsson, B.; Törnqvist, M. Cancer
risk from exposure to occupational acrylamide. Occup. EnViron.
Med. 2001, 58, 608-609.
(24) Friedman, M. A.; Dulak, L. H.; Stedman, M. A. A lifetime
oncogenicity study in rats with acrylamide. Fundam. Appl.
Toxicol. 1995, 284, 297-306.
(25) IARC. Acrylamide. In IARC Monographs on the EValuation of
Carcinogen Risk to Humans: Some Industrial Chemicals;
International Agency for Research on Cancer: Lyon, France,
1994; Vol. 60, pp 389-433.
(26) WHO. Guidelines for Drinking-Water Quality, 2nd ed.; World
Health Organization: Geneva, Switzerland, 1996; Vol. 2, pp
940-949.
(27) U.S. EPA. Assessment of Health Risks from Exposure to
Acrylamide; Office of Toxic Substances; U.S. Environmental
Protection Agency: Washington, DC, 1990.
(28) Segerbäck, D.; Calleman, C. J.; Schroeder, J. L.; Costa, L. G.;
Faustman, E. M. Formation of N-7-(2-carbamoyl-2-hydroxyeth-
yl)guanine in DNA of the mouse and the rat following intrap-
eritoneal administration of [14C]acrylamide. Carcinogenesis 1995,
16, 1161-1165.
(29) Paulsson, B.; Granath, F.; Grawé, J.; Ehrenberg, L.; Törnqvist,
M. The multiplicative model for cancer risk assessment: ap-
plicability to acrylamide. Carcinogenesis 2001, 22, 817-819,
1577.
(30) Paulsson, B.; Kotova, N.; Grawé, J.; Henderson, A.; Granath,
F.; Golding, B.; Törnqvist, M. Induction of micronuclei in mouse
and rat by glycidamide, the genotoxic metabolite of acrylamide.
Mutat. Res. 2002, submitted for publication.
(31) Granath, F. N.; Vaca, C. E.; Ehrenberg, L. G.; Törnqvist, M. A° .
Cancer risk estimation of genotoxic chemicals based on target
dose and a multiplicative model. Risk Anal. 1999, 19, 309-320.
(32) European Community. Council Directive 98/83/EC of Nov 3,
1998, on the quality of water intended for human consumption.
Off. J. Eur. Communities 1998, L 330 (05/12), 0032-0054.
(33) Swedish National Food Administration. http://www.slv.se.
(34) Rosén, J.; Hellenäs, K.-E. Analysis of acrylamide in cooked foods
by liquid chromatography tandem mass spectrometry. Analyst
2002, in press.
Received for review March 11, 2002. Revised manuscript received April
23, 2002. Accepted June 19, 2002. The work has been financially
supported by the Swedish Research Council for Environment, Agri-
cultural Science and Spatial Planning (FORMAS), the VL-foundation
R&D, Lidko1 ping, Sweden, the Swedish Association of Graduate
Engineers, and Stockholm University, faculty resources.
JF020302F